Good AI depends on old-fashioned wisdom
Recently, the Financial Times highlighted the uncomfortable discovery that 40% or so of European AI start-ups were not using AI.
This is hardly a surprise. Anyone who has worked in or around the tech industry for any time will know that whatever the latest 'disruptive' buzzword is, (including 'disruptive'), it will appear in a preposterously large number of company brochures and in an abundance of headlines in their marketing literature. The sad truth is, if you can hype up your company using the AI meme and if you can get venture capital funds more easily, you will always get some overstated claims.
But there is a flip side to this. If 40% of those surveyed abused the term, then shame on them. But remarkably, by implication, 60% of the 2,830 surveyed firms were using AI. That’s about 1,700 that passed from the sample, and the number is growing very rapidly. AI is out there. The wider report by MMC Ventures "The State of AI: Divergence” is very upbeat about this. In fact, it states that “AI may be the fastest major paradigm shift in the history of enterprise technology”. Enterprise technology is an interesting term and the reality is that many large enterprises are in the game, as well as the start-ups.
The shift to this rapid adoption is supported by the remarkable coincidence of several enablers.The adoption of machine learning, and reinforcement learning in particular, has been facilitated by platforms making this technique available to firms. 'Big data' exercises have made research data available to learning programs. Cloud infrastructure has made large computer capability available to do the processing needed and chip architecture has leaned itself to embed intelligent functions into hardware, such as natural language processing or visual recognition.
The existence of the set of tools to do the job is one thing. The lesson that 40% of AI firms don’t use AI is a warning to investors to do due diligence. In the same way, companies must do their due diligence on AI projects. This is particularly true for existing companies. Adopting the tools doesn’t make a successful project. Leaving aside that skills are scarce, even having a good project team is not sufficient. It is an inevitable truth that if you add technology without changing operating model, you must increase cost. Our CTO, Rens Troost, put out this same warning recently on robotic process automation. A model in which RPA appears to examine the legacy systems to integrate function or data into the new line is often fraught with unanticipated maintenance costs. What is the fastest growing area of AI according to the report? The answer is chat bots. Understanding the full consequential change of automation of conversations is going to be important and, with it, a new operating model around that. In turn, that demands a well-considered business case.
At Virtual Clarity, we regularly ask clients about their business case if we are invited in to help them change. In the example of AI, we would expect that business case to include the change in full operating model once the specific AI application is in operation. It forms a basis of working together to achieve the change and the results the client is looking for. Sometimes we have to help to put that in place. The case and its model provide strategic business context that allows something as exciting as AI to push through in the business. And the chances are, if you look at the range of case studies out there, the opportunity is worth the effort in the business case and the vision of the operational change.